* THE EXTERNAL POSSESSOR CONSTRUCTION IN WEST

Liliane Haegeman ([email protected])

1. AIM AND SCOPE OF THE PAPER

West Flemish (WF) has two constructions with prenominal possessors, illustrated in (1) and (2).

(1) a. Valère se boek b. Marie se boek Valère se book Marie se book 'Valère’s book' 'Marie’s book' c. Valère sen oto d. Marie sen oto Valère sen car Marie sen car 'Valère's car' 'Marie's car'

(2) a. Valère zenen boek Valère his book 'Valère’s book' b. Marie euren boek Marie her book 'Marie's book' c. Marie en Valère under boeken Marie and Valère their books 'Marie and Valère's books'

(1) illustrates the WF equivalent of the English or Dutch genitive construction. The bound morpheme se(n) has to be adjacent to the possessor DP. The alternation between se and sen is phonetically conditioned: se precedes a consonant, sen precedes a vowel. Se(n) is restricted to singular possessors. Though similar, and possibly diachronically related, to the masculine singular possessive pronoun zyn/zen, sen is used both for feminine and masculine possessors as shown by (1). In (2) the possessor DP is doubled by a possessive pronoun. The pronoun co-varies according to gender and number with the head noun.1 I refer to this construction as the doubling construction. The possessive pronoun in the doubling construction is a free

* I thank the editors of GG@G for inviting me to contribute a paper to this electronic journal. I hope the journal will become one of many ways to keep in touch with my friends @G. A first version of this paper was presented at the NP-conference in Antwerp (February 2000). I thank the audience of the conference for their comments. 1 Possessor doubling constructions have been signalled for a number of other languages. To mention just a few references: Ramat (1986) for a survey of such constructions in a range of languages, Corver (1990) for general discussion in a generative approach, Delsing (1998) and Fiva (1984) for Scandinavian, Jansen (1974, 1977), Janssen (1975) and Koelman (1975) for Dutch, Taeldeman (1995) for Flemish, Ponelis (1992) for , Stahl (1925) and Janda (1980) for earlier stages of English.

GG@G (Generative Grammar in Geneva) 1:1-19, 2000 © 2000 Liliane Haegeman 2 LILIANE HAEGEMAN morpheme and can be separated from the possessor (3a,a'); in the se(n) construction the possessor and the bound morpheme must be adjacent (3b,b').

(3) a. al Valère zen boeken all Valère his books a'. Valère al zen boeken b. al Valère-se boeken all Valère se books b'. * Valère al-se boeken

In this paper I will examine one property that sets the doubling construction apart from the genitive se(n) construction: the doubling construction allows for the pattern in (4a,b) in which an external relative or interrogative possessor is construed as having a possessor relation with a DP containing a matching doubling possessive pronoun. I will refer to this pattern as the external possessor construction.

(4) a. Dat is die verpleegster dan-ze gisteren [DP eur us] verkocht een. that is that nurse that-they yesterday her house sold have 'That’s the nurse whose house they sold yesterday.' b. Wekken verpleegster zei-je gie dan-ze gisteren [DP eur us] verkocht een? which nurse said you that-they yesterday her house sold have 'Who was the nurse whose house you said they sold yesterday?'

This pattern is not available for the sen-construction (5).

(5) a. * Dat is die verpleegster dan-ze gisteren [DP sen us] verkocht een. b. * Wekken verpleegster zei-je gie dan-ze gisteren [DP sen us] verkocht een?

Given the availability of the doubling construction in (2), it may at first sight seem tempting to propose to account for the data (4) in terms of (left-branch) extraction of the relative or interrogative possessor. The contrast between the grammaticality of (4) and the ungrammaticality of (5) follows if we assume that se(n) is a bound morpheme. Extraction of the possessor would lead to a violation of the Stray Affix filter. The same explanation would also account for the ungrammaticality of (3b'). This paper will provide evidence against a movement account of the data in (4). An analysis of (4) in terms of leftward possessor A'-movement would mean that WF is similar to Greek and Hungarian, in which possessor movement has been shown to be available:

(6) a. [CP [Tinosi ] [IP mu ipes [CP ti pos dhiavases [DP ti to vivlio]]] whose me-gen said that read-2sg the book 'You told me you read whose book?' (Greek, Horrocks and Stavrou 1987). b. [CP [TopP Marinaki [FocP PETER làtta [IP [DP ti a kalapja]]]]]. Mari-dative Peter saw the hat 'PETER saw Mary's hat.' c. [CP [FocP Kineki làtta [IP Kati [DP ti a kalapja]]]]? whose-dative saw Kati the hat 'Whose hat did Kati see?' (Hungarian, Szabolcsi 1983, 1987, 1994) THE EXTERNAL POSSESSOR CONSTRUCTION IN WEST FLEMISH 3

However, though attractive, such an analysis of the WF external possessor construction in (4) would lead to a range of unexpected asymmetries between WF and the related on the one hand and also internally to WF itself. The paper is organised as follows. Sections 2 and 3 offer arguments against a movement analysis for the WF external possessor construction. In section 4 I propose that the relation between the possessor and the possessed DP is established by the resumptive pronoun strategy and not by movement. Section 5 summarises the paper and contains some issues for future study.

2. GERMANIC ASYMMETRIES

Like WF, colloquial Dutch has a doubling pattern as discussed in some detail by Corver (1990).

(7) a. Du. Ze hebben die verpleegster haar/d'r huis verkocht. they have that nurse her/her house sold 'They have sold that nurse's house.' b. Du. Ik heb Jan z'n auto gewassen. I have Jan his car washed 'I have washed John's car.'

The external possessor construction is not available in Dutch, however:

(8) a. Du. * Dat is die verpleegster dat ze gisteren haar/d’r huis verkocht hebben. that is that nurse that they yesterday her house sold have b. Du. * Welke verpleegster zei jij dan ze gisteren haar/d’r huis verkocht hebben? which nurse said you that they yesterday her house sold have

Norwegian also displays the doubling construction (Fiva 1984, Corver 1990, Delsing 1998).

(9) Per sin bil (Norwegian: Fiva 1984; Delsing 1998) Per his car

As signalled by Corver (1990: 183), short possessor extraction is possible in Norwegian, but only if the possessed DP is a predicate (Compare the grammatical (10a,b,c) with the ungrammatical (10d,e)); long extraction is not possible.

(10) a. Hvemi er det [ti sin tante]? who is it his aunt 'Whose aunt is it?' b. Hvemi er det [ti sin bil]? who is it his car 'Whose car is this?' c. Hvemi er han [ti sin bror]? who is he his brother 'Whose brother is he?' d. * Hvemi skal vi forfore [ti sin soster] na? who shall we seduce his sister now e. *? Hvemi kjenner du [ti sin bror]? who know you his brother 4 LILIANE HAEGEMAN

Corver gives a case-related explanation for the Norwegian data. In order to account for the ban on extraction he says:

The idea that the specifier of a DP which is case marked by its sister is not accessible to antecedent government by a VP-adjoined trace, might shed some light on these puzzling facts. Consider first the facts in [10d,e], in which the possessor is reordered out of a DP which is not a predicate nominal. The verb assigns case to the DP containing the left-branch possessor. So it does not govern into the [Spec,DP] of that DP. The trace occupying the [Spec,DP] position after removal of the left-branch possessor, will not be antecedent governed by the VP-adjoined trace because of the barrierhood of the DP which is case marked by the verb. So, extraction yields an ECP- violation. (1990:183)

Extraction from a possessive construction is banned in Dutch, even from a predicative DP (11) (Corver 1990: 183).

(11) a. * Wiei is dat [ti z'n tante]? who is that his aunt b. * Wiei is hij [ti z'n broer]? who is he his brother

Corver relates the difference between Dutch and Norwegian to case assignment to the predicate DP. In Dutch, predicate nominals are assigned accusative case:

(12) a. Ik ben hem. I-nom am him-acc b. Ik denk that 't 'm is. I think that it him-acc is

Concerning examples like Dutch (12) he says:

Hem and its reduced form 'm are both predicate nominals carrying accusative case…the predicate nominal is assigned case by the verb, and does not, for example, receive nominative case under predication from the subject… The nonextractability of wie in [11] can be accounted for as follows: the copular verb assigns case to the DP- complement. Therefore, V cannot case-govern the [Spec,DP] position. Since V cannot govern the [Spec,DP] position, the intermediate antecedent trace adjoined to VP does not govern that position either. The case marked DP creates a barrier. So, the initial trace occupying the [Spec,DP] after removal of the left-branch possessor is not antecedent governed and therefore violates the ECP. (1990:184)

For Norwegian the situation is different since in that language the predicate pronoun may be realised as either nominative or accusative:

(13) a. Det er meg/jeg. that is me/I b. Det var ikke ham/han. that was not him/he THE EXTERNAL POSSESSOR CONSTRUCTION IN WEST FLEMISH 5

If the copular V does not assign case to the predicate, this means, according to Corver's view, that potentially the copula can govern the trace in [Spec,DP], rendering the trace accessible for an external antecedent-governor. Pursuing this pattern Corver (1990: 185) also considers the contrast in (14):

(14) a. [Hvem sin tante]i er det [ti sin katt]? who his aunt is that his cat b. * [Hvem]i er det [[ti sin tante ] sin katt]? who is that his aunt his cat

Extraction of the possessor hvem sin tante from the predicate [Spec,DP] in (14a) is grammatical; extraction of hvem from inside the possessive construction which is itself in [Spec,DP] is not possible in (14b). According to Corver, two problems arise in (14b). The extraction (14b) violates subjacency and it also violates the ECP: while the copular verb may be argued to govern the predicate DP and hence render the DP transparent according to Corver's analysis, this does not entail transparency of a more deeply embedded DP. The WF situation is markedly different from that in Norwegian. The examples in (4) show that a long distance relation between the possessor and the possessed DP is possible. These examples are not predicate constructions. Though short construal of an external possessor is possible, the WF equivalents of Norwegian (10) are actually slightly degraded (15).

2 (15) a. ?? Wieni is dat [ti zen tante]? who is that his aunt b. ?? Wieni is da [ti zenen oto]? who is that his car c. ?? Wieni ist-je [ti zen broere]? who is he his brother

Unfortunately, I will have no explanation to offer for the degradation of the examples in (15). I add here that increasing the distance between the possessor and the related DP improves the sentences:

(15) d. ? Wien was da doa gisteren [ti zenen oto]? who was that there yesterday his car e. ? Wieni is da gunter [ti zenen oto]? who is that there his car

Like Norwegian predicate pronouns, WF predicate pronouns may be nominative (16). However, this can hardly account for the productivity of the WF external possessor construction, since the WF external possessor construction is not restricted to predicate constructions and in fact it is slightly less favoured in predicate constructions than in non- predicate constructions.

(16) a. Wien was da? T was ik. who was that ? it was I b. Dat is zie niet. that is she not

2 In the representation I include a trace in order to represent the movement analysis. As we will see I reject this analysis later on. 6 LILIANE HAEGEMAN

Moreover, there is no degradation in WF when the external possessor is construed with a DP itself embedded inside the possessive [Spec,DP].

(17) a. Wien zei-je gie dan-ze [DP1 [DP2 [DP3 - ] zen dochter] eur us] a verkocht een? who said-you that-they [[[his daughter] her houses] already sold have b Dat is dienen vent dan-ze [DP1 [DP2 [DP3 -] zen dochter] eur us] a verkocht een. that is that man that-they his daughter her house already sold have

If the external possessor construction is to be analysed in terms of a left-branch A'- extraction, then the conclusion of this section has to be that WF differs from the other Germanic languages in allowing an unrestricted application of left-branch A'-extraction. In Standard Dutch possessor extraction is categorically excluded, and in Norwegian it is restricted to short extraction from predicate DPs. The asymmetries with the left-branch extraction data in the other Germanic languages at first sight cannot be made to follow from anything.

3. WF-INTERNAL ASYMMETRIES

This section shows that even if we restrict ourselves to a consideration of the WF data, a left- branch extraction analysis of the external possessor construction in (4) leads to a number of undesirable consequences. Among other things, such a movement analysis would lead us to conclude that WF wh-movement in the DP differs radically from wh-movement in the clause in that unlike wh-extraction from the clausal domain, wh-extraction from the DP is not restricted by any of the standard locality constraints on A'-movement identified in the literature (subjacency, wh- island, etc).

3.1. Extraction from a subject DP?

If the WF external possessor construction in (4) is derived by left-branch extraction, then such left-branch wh-extraction would also have to be allowed from DPs in the canonical subject position, i.e. the position to the immediate right of the (inflected) complementiser. (18) illustrates an active sentence, (19) a passive.

(18) a. Dat is dienen vent da [- zen broere] gisteren zen us verkocht eet. that is that man that [- his brother] yesterday his house sold has b. Wien zei-je gie da [- zen broere] gisteren zen us verkocht eet? who said-you that [- his brother] yesterday his house sold has

(19) a. Dat is dienen vent dan [- zen uzen] gisteren verkocht zyn. that is that man that his houses yesterday sold are b. Wien zei-je gie dan [- zen uzen] gisteren verkocht zyn? who said you that his houses yesterday sold are

WF definite subject DPs must be immediately adjacent to the complementiser, and nothing can intervene between the complementiser and the subject.3

3 Except for object clitics, which need not concern us here. THE EXTERNAL POSSESSOR CONSTRUCTION IN WEST FLEMISH 7

(20) a. Kpeinzen da (*gisteren) Valère zen us verkocht eet. I think that (*yesterday) Valère his house sold has 'I think that Valère sold his house yesterday.' b. Kpeinzen da (*gisteren) men broere zen us verkocht eet. I think that (*yesterday) my brother his house sold has 'I think that my brother sold his house yesterday.'

In (18) and (19), with possessive subjects, nothing can intervene between the complementiser da ('that') and the possessive subject DP, as shown by (21/22). It thus it seems reasonable to assume that the bracketed possessive subject DP in (18/19) occupies the canonical subject position.

(21) a. * Dat is dienen vent da gisteren [- zen broere] zen us verkocht eet. that is that man that yesterday [- his brother] his house sold has b. * Wien zei-je gie da gisteren [- zen broere] zen us verkocht eet? who said-you that yesterday [his brother] his house sold has

(22) a. * Dat is dienen vent dan gisteren [- zen uzen] verkocht zyn. that is that man that yesterday his houses sold are b. * Wien zei-je gie dan gisteren [- zen uzen] verkocht zyn? who said you that yesterday his houses sold are

If the separation of the relative/interrogative possessor from the possessed DP is the result of wh-extraction, we would have to conclude that WF freely allows left-branch extraction from the subject position. This result is surprising in the light of additional WF data, which I will discuss presently. First, cross-linguistically wh- extraction from the canonical subject position tends to be a marked phenomenon and often triggers language-particular mechanisms to ensure that the subject trace is licensed. In a pattern similar to the well-known que/qui alternation in French, WF extraction of a subject relative leads, obligatorily for some speakers and optionally for others, to an alternative form of the complementiser: da is replaced by die.

(23) a. Dat is dienen vent die/da dienen buot gekocht eet. that is that man who/that that boat bought has 'That is that man who bought the boat.' b. dat is dienen vent dan-ze zeggen die/da dienen buot gekocht eet. that is that man that-they say who/that that boat bought has 'That is the man who they say bought the boat.'

However, what would be analysed as an extraction of a left-branch possessor from inside a subject DP does not lead to such a da/die alternation:

(24) a. Dat is dienen vent [CP da/*die [IP [DP ti zen broere] gisteren zen us verkocht eet].] that is that man that his brother yesterday his house sold has b. Dat is dienen vent [CP dan/*dien [IP [DP ti zen uzen] gisteren verkocht zyn].] that is that man that his houses yesterday sold are

I cannot pursue in detail all the possible mechanisms that might be invoked for the licensing of the DP-internal trace, since much depends on the view one adopts of such mechanisms. Let 8 LILIANE HAEGEMAN me concentrate on the main issues. I assume, following Rizzi (1986, 1990), that a trace is subject to a dual requirement consisting of formal licensing and identification. The WF complementiser agrees in person and number with the subject (see Bennis and Haegeman 1984, Haegeman 1992):

(25) a. K peinzen da/*dan dienen student nen buot gekocht eet. I think that-3sg/*that-3pl that student a boat bought has b. K peinzen dan/*da die studenten nen buot gekocht een. I think that-3pl/*that-3sg those students a boat bought have

If the possessive subject DP did contain a trace of the external possessor, the agreeing complementiser might be argued to be implicated in the licensing of the trace. However, it seems unlikely that as such the complementiser can be held responsible for identification of an empty category in the (adjacent) specifier of the subject. We have already seen that there is no da/die alternation. Moreover, the complementiser agrees with the containing subject-DP and cannot agree with the possessor. In (26), for instance, the external possessor is singular (dienen vent, 'that man') but the complementiser agrees with the possessed plural DP (zen broers, 'his brothers').

(26) a. Kpeinzen dan/*da [[dienen vent] zen broers] nen buot gekocht een. I think that-pl/*that-sg that man his brothers a boat bought have 'I think that that man's brothers have bought a boat.' b. Dat is dienen vent dan [DP zen broers] nen buot gekocht een. that is that man that-3pl his brother a boat bought have 'that is the guy whose brothers bought a boat' c. * that is dienen vent da [DP zen broers] nen buot gekocht een that is that man that-3sg his brother a boat bought have

We might thus assume that the subject-internal trace of the possessive is fully licensed and identified within the possessive subject DP, perhaps by virtue of the presence of the doubling possessive pronoun. Recall that the doubling pronoun systematically agrees with the external possessor in terms of number, person and gender features. The pronoun could thus plausibly be argued to identify the trace. If indeed the WF possessor trace is taken to be licensed by virtue of the doubling pronoun, however, it is not at all clear what bans extraction in the related Germanic languages (Dutch, German, also Norwegian in most cases) with a similar doubling construction. In the previous examples I have concentrated on relativisation from a possessive subject DP in WF. Things become more intricate when we also take into consideration interrogative extraction. The data go against a movement analysis for the external possessor construction. Before providing the relevant argument we need to consider some of the properties of interrogative extraction in WF. Consider (27). Er-insertion is strictly obligatory with all indefinite (non-generic) subjects, regardless of whether the predicate is transitive or not.

(27) a. Kpeinzen dan *(der) vee studenten dienen boek goan kuopen. I think that *(there) many students that book go buy 'I think that many students will buy that book.' b. Kpeinzen dan *(der) drie uzen verkocht zyn. I think that there three houses sold are 'I think that three houses have been sold.' THE EXTERNAL POSSESSOR CONSTRUCTION IN WEST FLEMISH 9

c. Kpeinzen dan *(der) studenten no de feeste goan kommen. I think that *(there) students to the party go come 'I think that students will come to the party.'

An indefinite subject4, unlike a definite subject DP, can be argued to occupy a position which is lower than the canonical subject position. As shown by (28), an adjunct can intervene between the complementiser + er and the indefinite subject:

(28) a. Kpeinzen dan *(der) morgen vee studenten dienen boek goan kuopen. I think that *(there) tomorrow many students that book go buy 'I think that many students will buy that book tomorrow.' b. Kpeinzen dan *(der) gisteren drie uzen verkocht zyn. I think that there yesterday three houses sold are 'I think that three houses were sold yesterday.' c. Kpeinzen dan *(der) morgen studenten no de feeste goan kommen. I think that *(there) tomorrow students to the party go come 'I think that students will come to the party tomorrow.'

In WF, extraction of the interrogative subject pronoun wien ('who') always triggers er- insertion.

(29) a. Wien peinzje-gie dat *(der) dat us goa kopen? who think you that *(there) that house goes buy 'Who do you think will buy that house?'

This suggests that interrogative pronouns and their traces have the syntactic status of indefinites DPs. Like indefinites they are banned from the canonical C-adjacent subject position and occupy a lower position. Interrogative wh-extraction is not launched from the canonical subject position but from a lower position. As a result, we also do not expect there to be a need for the da/die alternation.

(29) b. * Wien peinzj-gie die ter dat us goa kuopen?

What we might call the indefiniteness effect for interrogative subjects is also present with D- linked interrogative subjects such as wekken N ('which N') or wekeenen van die N ('which one of those N').

(29) c. Weknen zeune peinzdeg-je gie dat *(der) dat us ging kopen? which son thought-you you that *(there) that house went buy 'Which son did you think was going to buy that house?' d. Wekeenen van die zeuns peinzdeg-je gie dat *(der) dat us ging kopen which one of those sons thought-you you that *(there) that house went buy 'Which one of these sons did you think was going to buy that house?'

DPs may be indefinite due to a number of factors. One is the presence of an indefinite article, a quantifier or a numeral. Also indefinite are DPs with a prenominal indefinite possessor. This is shown by the availability of the there construction in English (30a) in

4 For the discussion I treat quantified DPs as indefinites. There are obviously differences but these do not bear on the present argument. 10 LILIANE HAEGEMAN which the DP in construction with there contains an indefinite possessor, a student's. (30a) constrasts with (30b) with the definite possessor:

(30) a. There is a student's mother waiting in your office. b. * There is the new student's mother waiting in your office.

In WF too, the presence of a prenominal indefinite possessor DP in the doubling construction renders a DP indefinite. Like all indefinite DPs, the relevant DP requires er-insertion when a subject (31a). A prenominal definite possessor DP does not have this effect (31b).

(31) a. Kpeinzen dat *(ter) [nen student zen moeder] da geschreven eet. I think that *(there) a student his mother that written has 'I think that a student's mother has written this' b. K peinzen dat (*ter) [dienen student zen moeder] da geschreven eet. I think that (*there) that student his mother that written has 'I think that that student's mother has written that.'

For completeness' sake note that the presence of a prenominal indefinite possessor with se(n) (see (1) above) also renders the containing DP indefinite:

(32) Kpeinzen dat *(ter) [nen student se moeder] da geschreven eet. I think that *(there) [a student se mother] that written has 'I think that a student's mother has written this'

A possessive subject DP with an indefinite prenominal possessor can be separated from the complementiser by an intervening maximal projection (33a,b), but a possessive subject DP with a definite pronominal possessor must be adjacent to the complementiser (33c).

(33) a. Kpeinzen datter tun [nen student zen moeder] da geschreven eet. I think that there then a student his mother that written has 'I think that a student's mother then wrote this.' b. Kpeinzen datter tun [nen student se moeder] da geschreven eet. I think that there then a student se mother that written has 'I think that a student's mother then wrote this.' c. * Kpeinzen dat tun [dienen student zen moeder] da geschreven eet. I think that then that student his mother that written has

Negative quantifiers such as niemand ('no one') behave syntactically as indefinites, requiring er-insertion when subjects (34a) and allowing for the occurrence of a constituent separating them from the complementiser.5 In (34b), for instance, the direct object dienen boek ( 'that book') precedes the negative subject.

(34) a. Kpeinzen dat *(ter) niemand dienen boek kent. I think that *(there) no one that book knows 'I think that no one knows that book.' b. Kpeinzen dat *(ter) dienen boek niemand kent. I think that *(there) that book no one knows

5 For differences between N-words and indefinites see Haegeman (1997). THE EXTERNAL POSSESSOR CONSTRUCTION IN WEST FLEMISH 11

As expected, the presence of a prenominal possessive, niemand, turns a possessive DP into an indefinite. When such a DP is a subject there is obligatory er-insertion and a constituent may intervene between the DP and the complementiser + er.

(35) a. Kpeinzen da *(ter) [niemand zen oukders] da weten. I think that *(there) no one his parents that know 'I think that no one's parents know that.' b Kpeinzen da *(ter) da neu nog [niemand zen oukders] weten. I think that *(there) that now no one his parents know 'I think that no one's parents know it yet.'

Let us now return to the external possessor construction in (4). A left-branch extraction analysis for the examples in WF (4) raises a problem in the light of the discussion above in which the following points have been established.

(i) Interrogative DPs such as wien act syntactically as indefinites. They require er- insertion when they are subjects and the extraction site is not the canonical subject position. (ii) The presence of a prenominal indefinite possessor DP turns the containing DP into an indefinite.

If the external possessor construction in (4) is derived by extraction then the possessed DP must contain a trace of the external possessor. A subject DP containing the trace of the external possessor wien will be expected to be indefinite. In other words, we expect such a possessive subject DP (i) not to be able to occupy the canonical subject position, which is strictly reserved for definites; (ii) to require er-insertion; (iii) to be able to be separated from the complementiser position by intervening material. These predictions are not borne out. As shown by (36a), the possessive subject DP which, under an extraction account, should contain the trace of wien is incompatible with er- insertion. As shown by (36b), the possessive subject DP must be adjacent to the complementiser.

(36) a. [Wien] zei-je gie [CP da (*der) [DP ti zen broere] zen us verkocht eet]? who said you that (*there) his brother his house sold has 'Whose brother did you say had sold his house?' b. * [Wien] zei-je gie [CP da verleden joare [DP ti zen broere] zen us verkocht eet]? who said you that last year his brother his house sold has

Replacing wien by a D-linked interrogative will not alter the judgements:

(37) a. [Weknen zeune] zei-je gie [CP da (*der) [DP ti zen wuf] nen oto gekocht eet]? which son said you that (*there) his wife a car bought has 'Which son's wife did you say has bought a new car?' b. * [Weknen zeune] zei-je gie [CP da tun [DP ti zen wuf] nen oto gekocht eet]? which son said you that then his wife a new car bought has 12 LILIANE HAEGEMAN

We are led to the conclusion that the possessed DP which is construed with an external wh- interrogative possessor does not behave as if it contained a trace of indefinite interrogative wien. Rather, the DP behaves as if it were syntactically definite. It is difficult to see how to account for the difference between wh-extraction of argument wien, whose trace is syntactically indefinite, and wh-extraction of the prenominal possessor wien, whose trace would have to be definite. This asymmetry casts further doubts on a left-branch extraction/movement analysis in which an external possessor is construed with a trace. The next section offers further objections to the movement analysis.

3.2. Wh-islands

Under a movement analysis the WF external possessor construction in (4) would be derived by long extraction of possessors. Another undesirable consequence of such a movement analysis is that unbounded movement would be required. As shown by (38), the external possessor can be construed with a possessive DP contained in a wh-island.

(38) a. Dat is dienen vent dank nie weten [of da zen moeder tun hertrouwd is]. that is that man that-I don't know if that his mother then remarried is b. Dat is dienen vent dank nie weten [me wien da zen moeder hertrouwd is]. that is that man that-I don't remember with whom that his mother remarried is c. Dat is dienen vent dank nie weten [of dan-ze zen moeder a gevroagd eet]. that is that man that I not know if that-they his mother already asked has d. Dat is dienen vent dank nie weten [wien dat-er zen moeder gevroagd eet]. that is that man that-I not know who that there his mother asked has

Normally, wh-extraction of a clausal constituent in WF is subject to wh-island constraints as shown by the degradations in (39).

(39) a. ?? Dat is dienen schryver dank nie weten [of dan de studenten t kennen]. that is that writer that I not know whether that the students know b. ?? Dat is dienen vent dan-k nie verstoan [woarom dan men ukders t nie vroagen]. that is that man whom I not understand why that my parents not invite

Furthermore, WF displays an asymmetry between object extraction from an island, which is marginal (39), and subject extraction, which is ungrammatical (40).

(40) a. * dat is dienen schryver dank nie weten [of da t men studenten kent]. that is that author that-I not know whether that my students knows b. * dat is dienen vent dank nie verstoan [woarom da t men oukders nie vroagt]. that is that man that-I not understand why that my parents not invites

The external possessor, on the other hand, can be construed with a possessed DP across an island, regardless whether the relevant DP is subject (38a,b) or object (38c,d). To overcome island violations as those in (39) and (40), WF uses resumptive pronouns.

(41) a. Dat is dienen schryver dank nie weten [of dan de studenten em kennen]. that is that writer that I not know whether that the students him know b. Dat is dienen vent dan-k nie verstoan [woarom dan men ukders em nie vroagen]. that is that man whom I not understand why that my parents him not invite THE EXTERNAL POSSESSOR CONSTRUCTION IN WEST FLEMISH 13

(42) a. Dat is dienen schryver dank nie weten [of datje men studenten kent]. that is that author that-I not know whether that-he my students knows b. Dat is dienen vent dank nie verstoan [woarom datje men oukders nie vroagt]. that is that man that-I not understand why that-he my parents not invites

The data in (41/42) hold the key to an analysis of the external possessor, as we shall see presently. First, I will provide additional arguments against the movement analysis.

3.3. Recursive possessives and left-branch extraction

As mentioned, the very fact that the WF construction would display a left-branch extraction is remarkable since such extractions are severely limited cross-linguistically (Corver 1990). The data in (17) above and in (43) below show that in the external possessor construction we would have to postulate that a left-branch can be extracted from a specifier position which itself is more deeply embedded inside another possessive specifier. Consider (43a): dienen vent ('that man') is the prenominal possessor inside DP2 dienen vent zen dochter ('that man his daughter'), which, in turn, is the prenominal possessor for DP1 dienen vent zen dochter eur us.

(43) a. Z'een [DP1 [DP2 [DP3 dienen vent] zen dochter] eur us] a verkocht. they have that man his daughter her house already sold

As shown by (17), repeated here as (43b,c), interrogative extraction or relativisation on the basis of the most deeply embedded left-branch dienen vent is possible:

(43) b. Wien zei-je gie dan-ze [DP1 [DP2 [DP3 - ] zen dochter] eur us] a verkocht een? who said-you that-they [[[his daughter] her house] already sold have c. Dat is dienen vent dan-ze [DP1 [DP2 [DP3 -] zen dochter] eur us] a verkocht een. that is that man that-they his daughter her house already sold have

A movement analysis would entail that the external possessor in WF can cross two DP boundaries without any problem. For completeness' sake, note that the recursive pattern of left-branches can also be instantiated with the sen construction (44a). As shown by (44b,c) the two construction types may co-occur in the same DP.

(44) a. Z'een [DP1 [DP2 [DP3 Marie] se dochter] sen us] a verkocht. they have Marie se daughter sen house already sold b. Z'een [DP1 [DP2 [DP3 Marie] se dochter] eur us] a verkocht. they have Marie se daughter her house already sold c. Z'een [DP1 [DP2 [DP3 Marie] eur dochter] sen us] a verkocht they have Marie her daughter sen house already sold

The doubling construction can be construed with a possessor external to the DP, the sen construction cannot:

(45) a. * Wien zei-je gie dan-ze [DP1 [DP2 [DP3 - ] se dochter] eur us] a verkocht een? who said-you that-they [[[se daughter] her house] already sold have b. * Dat is dienen vent dan-ze [DP1 [DP2 [DP3 -] se dochter] eur us] a verkocht een. that is that man that-they se daughter her house already sold have 14 LILIANE HAEGEMAN

An external interrogative or relative possessor can be related to a possessive DP with doubling possessive pronoun, which itself is the possessor in a sen construction:

(46) a. Wien zei-je gie dan-ze [DP1 [DP2 [DP3 - ] eur dochter] sen us] a verkocht een? who said-you that-they [[- her daughter] sen house] already sold have b. Dat is die moeder dan-ze [DP1 [DP2 [DP3 -] eur dochter] sen us] a verkocht een. that is that mother that-they [[-her daughter] sen house] already sold have

An analysis of the external possessor in terms of an unconstrained left-branch extraction in the WF possessive constructions would mean that this kind of extraction contrasts systematically with wh-extraction in the clausal domain. This asymmetry between the application of wh-movement in the clause and that in the DP goes against the trend that attempts to bring the properties of DP in line with these of the clause.

3.4. Extraction from extraposed PPs

In WF, as in standard Dutch and German, wh-extraction from a PP (i.e. with stranding) is only possible from the middle field. In (47a) an R-pronoun is extracted from the PP in the middle field, stranding the preposition an ('on'). The PP an die zoake ('on that matter') can also appear in extraposed position (47b), but in that case P-stranding is not possible (47c).

(47) a. Woari eej-gie gisteren [PP ti an] gewerkt? where have you yesterday on worked 'What did you work on yesterday?' b. K'een gisteren gewerkt an die zoake. I have yesterday worked on that matter. c. * Woari eej-gie gisteren gewerkt [PP ti an]?

The syntax of extraposed constituents awaits a definitive analysis6. In traditional GB-terms, the ungrammaticality of the extraction of woar in (47c) tends to be interpreted in terms of an ECP violation. Pursuing a formal licensing and identification analysis of ECP, we might assume that the PP itself does not contain the structure to both license and identify the trace and that the middle field position of the PP, though not the extraposed position, allows the trace to meet its licencing requirements. Observe, now, that in the external possessor construction it is perfectly possible to construe an external possessor with a possessive DP contained in an extraposed PP:

(48) a Wieni zei-je gie da-j vroeger nog gewerkt eet [PP vu [DP ti zenen zeune]]? who said you that you before worked have for his son 'For whose son did you say that you used to work?' b. Dat is dienen venti dank vroeger nog gewerkt een [PP vu [DP ti zenen zeune]]? that is that man that I before worked have for his son 'That is the guy whose son I used to work for.'

If we assume a movement analysis, we are once again led to the conclusion that the WF possessor DP would have to contain the structure and properties required to license the left- branch trace. It is not clear, again, how the language could then be differentiated from Standard Dutch and German, which do not allow for such constructions.

6 See Haegeman (1998, 1999) for some discussion on extraposition in WF. THE EXTERNAL POSSESSOR CONSTRUCTION IN WEST FLEMISH 15

4. AN ALTERNATIVE PROPOSAL: RESUMPTIVE PRONOUNS

In the preceding sections we have examined some aspects of the syntax of the external possessor construction illustrated in (4), repeated here as (49).

(49) a. Dat is die verpleegster dan-ze gisteren [eur us] verkocht een. that is that nurse that-they yesterday [her house] sold have 'That’s the nurse whose house they sold yesterday.' b. Wekken verpleegster zei-je gie dan-ze gisteren [eur us] verkocht een? which nurse said you that-they yesterday [her house] sold have 'Who was the nurse whose house you said they sold yesterday?'

An analysis in which the external wh-possessor - relative in (49a) and interrogative in (49b) – is moved from the left periphery of the DP to the left periphery of the clause might at first sight have seemed the most attractive path to pursue. Such an approach would make WF interestingly similar to other languages with possessor extraction and would reveal an unexpected contrast with the Germanic languages. However, the overview of related constructions in WF and in Germanic languages shows that apart from the obvious problems independently associated with left-branch extractions (Corver 1990), a movement analysis, for the external possessor construction raises a range of problems. In essence, what would have to be left-branch extraction in the WF doubling construction would have to be associated with a set of properties radically distinct from the extraction patterns in the language in general – i.e. outside the possessive domain - and from the similar doubling constructions in Dutch and German and in Scandinavian. I summarise these major asymmetries here.

(i) WF would be different from its German and Dutch counterparts in being the only West Germanic language to freely allow possessor extraction (i.e. a left-branch extraction) in the doubling construction. (ii) The external possessor in the WF doubling construction fails to display any of the constraints on possessor extraction found with Norwegian doubling. (iii) While WF displays constraints on extractions from the canonical subject position; left- branch possessor extraction from inside that position would have to be completely unconstrained. (iv) In WF, the trace of a subject interrogative wh-constituent has the features of an indefinite DP in that it always triggers er-insertion. The alleged trace of the external interrogative wh-possessor, on the other hand, would unexpectedly have to be definite. (v) The WF external possessor construction displays no degradation – i.e. neither a strong ECP effect nor a weaker subjacency effect - when an additional DP boundary is crossed. (vi) While WF exhibits wh-island effects in relative and interrogative A'-movement, possessor movement would have to be unconstrained. (vii) The subject/object asymmetries normally found with WF wh-extraction from wh-islands would have to be absent in the case of possessor extraction. (viii) Left-branch extraction of the possessor would also have to apply to DPs contained in extraposed PPs, which normally do not allow extraction

An A'-movement analysis of the external possessor construction leads to a representation as in (50).

(50) a. [CP whi … [DP ti POSS … wien zenen 16 LILIANE HAEGEMAN

If we maintain that the external possessor construction is derived by movement, we will have to complicate the grammar of WF significantly in order to accommodate all the asymmetries listed above. To avoid such complications I will reject the movement analysis and I propose that the external possessor is related to the possessed DP by virtue of construal with a resumptive pronoun. There are a number of distinct ways of implementing this idea. One option is to assume that the doubling possessive pronoun itself is the resumptive pronoun. Alternatively, given the clitic nature of the possessor zenen, which contrasts with a stronger form zynen, it could be proposed that the possessive pronoun is a clitic, i.e occupies a head position and that it identifies a non-overt pronominal, pro, in a specifier-position. As far as I can tell, the data discussed so far will not provide us with any particular information as to the precise position of the non-overt pronoun. Following various discussion in the literature (Zribi Hertz 1998, Picallo 1994) we could assume that the clitic possessive, whose presence is incompatible with the determiner, cliticises to D. Let us assume provisionally that [Spec, DP] would then be occupied by pro. On the basis of these assumptions, (50a) is replaced by (50b), in which an operator in a left-peripheral position, here the interrogative pronoun wien, binds the null pronoun licensed by the clitic possessor zenen.7

(50) b. [CP whi … [DP proi [D POSS] … ]] wien zenen

On the basis of (50b), the external possessor construction is no longer singled out as having a range of special properties. External possessor constructions are instantiations of construal of a wh-operator with a resumptive pronoun (51a). Resumptive pronouns are independently available in WF as shown by (41/42) above and by (51b,c).

(51) a. Wieni ee-j gie men neu a were gevroagd of da [proi zen moeder] nog leeft? who have-you you me now already asked whether that his mother still lives b. Wavuonen studenti ee-j gie men neu a were gevroagd of da tjei getrowd was? which student have-you you me now already asked whether that he married was c. Dat is dienen venti dan-k nie mie weten of dat-jei getrouwd is that is that man that-I now no more know whether that he married is

Like all pronouns, the DP-internal resumptive pronoun construed with the external possessor will be interpreted as definite (52a).8 Note that a subject pronoun bound by a quantificational element continues to occupy de canonical subject position (52b,c):

7 It is not obvious that pro occupies [Spec,DP].Consider the pair (ia,b). In (ia) the possessor Jan precedes the possessive pronoun, in (ib) the head N of the possessed DP is ellipted and the possessive pronoun is realised in its full form, the strong pronoun zynen. Presumably the strong form of the pronoun is required to licence the ellipsis (cf. Picallo 1995). The strong pronoun follows the determiner de. (i) a. Jan zen boeken Jan his books b. Jan de zyne Jan the his We might speculate that in (ia) the cliticisation of zenen to D licences pro in the lower position occupied by zynen in (ib). c. Jan zen [pro boeken] 8 In WF the occurrence of the full pronoun zie in post-complementiser position without a doubling clitic is marked. The more natural forms would be: (i) a. Kweten nie of da-se zie getrowd is. I know not if that -she she married is THE EXTERNAL POSSESSOR CONSTRUCTION IN WEST FLEMISH 17

(52) a. Kweten nie of da (*der) zie getrowd is. I know not if that (*there) she married is. b. Iedere studente peinst da (*er) zie de beste keure eet. Every student thinks that (*there) she the best chance has. c. Wekken studente peinster niet da (*er) zie nog tyd eet? which student thinks-there not that (*there) she still time has

A possessive DP containing a (definite) resumptive possessive pronoun as its specifier will have the properties of a definite DP. Hence, when such a DP is a subject, it will, like all definite subject DPs, occupy the canonical subject position. This means that er-insertion will not be possible and the subject DP has to be adjacent to the complementiser.

(53) a. Kweten nie of da (*er) zen moeder nog leeft. I know not if that (*there) his mother still lives b. Kweten nie of da (*gisteren) zen moeder nog leefde. I know not if that (*gisteren) his mother still lived c. Wieni eej-gie men neu a were gevroagd of da (*er) zen moeder nog leeft? who have you me now already asked whether that (*there) his mother still lives d. Wieni eej-gie men a were gevroagd of da (*gisteren) zen moeder nog leefde? who have you me already asked whether that (*yesterday) his mother still lived

5. WF AND STANDARD DUTCH

Obviously we need to address the question what distinguishes standard Dutch, which lacks the external possessor construction in (4), from WF which has it productively. An answer to this question rests on the observation that the resumptive pronoun strategy in general is not available in standard Dutch. The absence of the resumptive pronoun strategy will entail correctly that standard Dutch lacks the equivalent of (3a'), repeated here as (54b) in which the possessor DP is not adjacent to the possessive pronoun.

(54) a. al Valère zen boeken all Valère his books b. Valère al zen boeken

(55) a. al Jan z'n boeken(Dutch) all Jan his books b. * Jan al z'n boeken(Dutch)

b. Kweten nie of da-se getrowd is. I know not if that -she married is Again, in such cases, er-insertion is not possible. Observe that as such the presence of the clitic se need not rule out the presence of er, as long as this is not an instantiation of the existential er. Partitive (d)er is for instance possible. (ii) a. Kweten nie of da-se zie der vele kent I know not if that -she she there many knows 'I don't know if she knows many.' b. Kweten nie of da-se der zie vele kent I know not if that -she there she many knows 18 LILIANE HAEGEMAN

The absence of resumptive pronouns in Dutch and its availability in WF may also be related to subject pronoun doubling in WF. As I have discussed in earlier work (Haegeman 1992), WF full subject pronouns are often doubled by a clitic. This strategy is not available in standard Dutch.

(56) a. ze goa zie no Gent she goes she to Ghent b. ze goan zunder no Gent they go they to Ghent

6. CONCLUSION AND QUESTIONS FOR FUTURE STUDY

In this paper I have discussed the external possessor construction displayed in WF (4) repeated here as (57).

(57) a. Dat is die verpleegster dan-ze gisteren eur us verkocht een. that is that nurse that-they yesterday her house sold have 'That’s the nurse whose house they sold yesterday' b. Wekken verpleegster zei-je gie dan-ze gisteren eur us verkocht een? which nurse said you that-they yesterday her house sold have 'Who was the nurse whose house you said they sold yesterday?'

One might be tempted to derive (57) by means of a movement analysis in which a left-branch possessor is extracted from a doubling possessor construction. This paper shows that such an analysis is not very plausible as the movement invoked to derive (57) would be subject to none of the constraints associated with movement in WF. I propose therefore that the external possessor is related to the possessive DP by being construed with a resumptive pronoun. I have offered two implementations of the resumptive pronoun analysis. In future work I intend to try to narrow down these options in order to shed more light on the left-periphery of the DP (see also note 7).

REFERENCES

Alexiadou, A. and C. Wilder (eds.) (1998) Possessors, Predicates and Movement in the DP, Benjamins, Amsterdam. Bennis, H. and L. Haegeman (1984) "On the Status of Agreement: COMP and INFL in Flemish Dialects", in W. de Geest and Y. Putseys (eds.) Sentential Complementation, Foris, Dordrecht, 33-53. Corver, N. (1990) The Syntax of Left Branch Extractions, Ph.D. Dissertation, Tilburg. Delsing, L.-O. (1998) "Possession in Germanic", in Alexiadou and Wilder (eds.), 87-108. Fiva, T. (1984) "NP-Internal Chains in Norwegian", Working Papers in Scandinavian Syntax, 14. Haegeman, L. (1992). Theory and Description in Generative Grammar. A Case Study in West Flemish, Cambridge University Press, Cambridge. Haegeman, L. (1997) "N-Words, Indefinites and the Neg Criterion", in D. Forget, P. Hirschbühler, F. Martineau and M.-L. Rivero (eds.) Negation and Polarity. Syntax and Semantics, Benjamins, Amsterdam, 115-137. THE EXTERNAL POSSESSOR CONSTRUCTION IN WEST FLEMISH 19

Haegeman, L. (1998) "'Extraposed' Clauses in the West Germanic SOV Languages", in J. D. Fodor and S. J. Keyser (eds.) Chomsky Celebration Website, MIT Press. Also published in GenGenP 6.1., 61-70. Horrocks, G. and M. Stavrou (1987) "Bounding Theory and Greek Syntax: Evidence for Wh- Movement in NP", Journal of Linguistics 23, 79-108. Janda, R. (1980) "On the Decline of Declensional Systems: The Overall Loss of OE Nominal Case Inflections and the ME Reanalysis of –es as his", in E. Traugott et al. (eds.) Papers from the Fourth International Conference on Historical Linguistics, Benjamins, Amsterdam, 243-252. Jansen, F. (1974) "Wat is Jan z'n boek?", Tabu 1-2, 17-19. Jansen, F. (1977) "Jan z'n boek, een genitief?" Spektator 7-8, 436-40. Janssen, Theo (1975) "Possessieve konstrukties", De Nieuwe Taalgids 68, 1-13. Koelman, L. (1975) "Jan z'n boek en de pre-genitieven", De Nieuwe Taalgids 68, 433-45. Picallo, C. (1994) "Catalan Possessive Pronouns: The Avoid Pronoun Principle", Natural Language and Linguistic Theory 12, 259-299. Ponelis, F. A (1992) "Die ontwikkeling van die Possessief in Afrikaans", in H. Bennis and J.W. de Vries (eds.), De binnenbouw van het Nederlands, ICG publications, Dordrecht, 273-290. Ramat, P. (1986) "The Germanic Possessive Type: dem Vater sein Haus", in D. Kastovsky and A. Szwedek (eds.) Linguistics Across Historical and Geographical Boundaries. Trends in Linguistics, Studies and Monographs 32, Mouton de Gruyter, Berlin, 579- 590. Rizzi, Luigi (1986) "Null Objects in Italian and the Theory of pro", Linguistic Inquiry 17, 501-557. Rizzi, L. (1990) Relativized Minimality, MIT Press, Cambridge. Stahl, L. (1925) Der adnominale Genitiv und sein Ersatz im Mittelenglischen und Frühneuenglischen. Selbstverlag des Englischen Seminars der Universität Giessen, Giessen. Szabolcsi, A. (1983) "The Possessor that Ran Away from Home", Linguistic Review 1, 89- 102. Szabolcsi, A. (1994) "The Noun Phrase", Syntax and Semantics 27. The Syntactic Structure of Hungarian, Academic Press, New York, 179-274. Taeldeman, J. (1995) "Jan Z'N + nomen. Over een bezitsconstructie in de Vlaamse dialecten", Taal en Tongval 47, 220-228. Zribi-Hertz, A. (1998) "Les syntagmes nominaux possessifs en français moderne: syntaxe et morphologie", in J. Guéron and A. Zribi-Hertz (eds.), La grammaire de la possession, Université Paris X, Nanterre, 129-193.

Author's affiliation UFR Angellier UMR 8528 Silex du CNRS Université Charles de Gaulle III